[Solved] Installing Ethernet Drivers on Installation Media

[NeddySeagoon]

xevra,

Your /boot is good. What of your grub.cfg?
That's in /boot/grub/

[xevra]

I suppose you do have initrd support enabled in the kernel?

NeddySeagoon,

I had "ugrd' enabled in the USE flags, so the initramfs should be enabled in the kernel. The distribution kernel requires that you use one or the other.

On my other machine, I successfully booted Gentoo using a distribution kernel compiled with the same flags.

Additionally, when I try to emerge the distribution kernel again, with "--newuse", it says there is nothing to be done.

However, it's always better to put eyes on it than trust that I executed the same command on both machines. Is there a way for me to explicitly check?

They should both be in the grub menu

The grub menu presently suggests:

Code: Select all

Gentoo GNU/Linux
Advanced options for Gentoo GNU/Linux
UEFI Firmware Settings

I only see one Gentoo listing, and it's not specific enough for me to identify it as one or the other of those files.

One boot stanza in grub.cfg should say

Code: Select all

echo "Loading kernel ..."
kernel vmlinuz-6.13.4-gentoo-dist ...
echo "Loading initrd ..." 
initrd initramfs-6.13.4-gentoo-dist.img

After running a fresh grub-mkconfig, it looks like this: pastebin

[NeddySeagoon]

xevra,

That's good, almost

Code: Select all

	echo	'Loading Linux 6.13.4-gentoo-dist ...'
	linux	/boot/vmlinuz-6.13.4-gentoo-dist root=UUID=4643c03f-299a-4e28-b975-5c95b56244f0 ro net.ifnames=0 
	echo	'Loading initial ramdisk ...'
	initrd	/boot/amd-uc.img /boot/initramfs-6.13.4-gentoo-dist.img

I didn't see /boot/amd-uc.img in your /boot. I don't knew if that's a problem, with it missing.
As I don't use grub, I don't know how it got autodetected unless it was in /boot

Do you have

Code: Select all

CONFIG_BLK_DEV_INITRD=y

set in your kernel?
The kernel will not use the initrd without it.

[xevra]

xevra,

That's good, almost

Code: Select all

	echo	'Loading Linux 6.13.4-gentoo-dist ...'
	linux	/boot/vmlinuz-6.13.4-gentoo-dist root=UUID=4643c03f-299a-4e28-b975-5c95b56244f0 ro net.ifnames=0 
	echo	'Loading initial ramdisk ...'
	initrd	/boot/amd-uc.img /boot/initramfs-6.13.4-gentoo-dist.img

I didn't see /boot/amd-uc.img in your /boot. I don't knew if that's a problem, with it missing.
As I don't use grub, I don't know how it got autodetected unless it was in /boot

Do you have

Code: Select all

CONFIG_BLK_DEV_INITRD=y

set in your kernel?
The kernel will not use the initrd without it.

NeddySeagoon,

I can confirm that CONFIG_BLK_DEV_INITRD=y is set in /boot/config-6.13.4-gentoo-dist

[NeddySeagoon]

xevra,

Does /boot/amd-uc.img exist ?

[xevra]

However I like to point out your SHOULD NOT get same kernel panic. At least the line that said

VFS: Cannot open root device "UUID=4643c03f-299a-4e28-b975-5c95b56244f0" or unknown-block(0,0)

would have changed to

VFS: Cannot open root device "PARTUUID=6f7921b9-6801-4905-9a33-506de5659ba4" or unknown-block(0,0)

I know I may be picky but since I am not set in front of the computer I have no way to judge what happen.

So if really come out exact same error messages then it mean your change to grub.cfg did not take effect, or you pick a wrong line to boot.

If PARTUUID=... still does not work then most likely your kernel have some problem detect nvme device.

Earlier, I was able to get a kernel panic method with the PARTUUID, and was still getting unknown-block(0,0), -6 errors, but "PARTUUID" was in the error.

I'm not sure I had done that correctly however, because while I could find and replace UUID=4643... with PARTUUID=6f... there's two other kinds of places where the UUID shows up. There's one kind of place where it's just part of and address, and another kind of place where it is in an --fs-uuid. At one point, I had tried changing that to --pt-uuid and replacing the value, following this, but I'm not sure that worked either. I'm not sure this person's work was ever merged upstream.

Following this, I ran another grub-mkconfig to reset what I had done, because it didn't seem right.

Edit: I don't know why I said "null-device." I meant "unknown-block"

[xevra]

xevra,

Does /boot/amd-uc.img exist ?

Yes, it is 145Kb, according to ls.

[NeddySeagoon]

xevra,

The numbers in

Code: Select all

unknown-block(0,0)

here (0,0) are the kernel major and minor device numbers.
(0,0) means that the kernel cannot find any block devices at all.

That's because the initrd didn't do anything.
Is

Code: Select all

CONFIG_BINFMT_SCRIPT=y

set?

The initrd is a root filesystem in a file. The kernel mounts it as root then executes the script called init.
The above option is required to permit scripts to execute.

[xevra]

NeddySeagoon,

Thanks for the explanation.

I can confirm that option is set.

Thank you,
Xevra

[NeddySeagoon]

xevra,

There is a list of devices at /usr/src/linux/Documentation/admin-guide/devices.txt

[xevra]

xevra,

There is a list of devices at /usr/src/linux/Documentation/admin-guide/devices.txt

That's a negative. There is a /usr/src/linux/Documentation folder, but there is no admin-guide within that folder.

[NeddySeagoon]

xevra,

It's in git too https://github.com/torvalds/linux/blob/ ... evices.txt

[pietinger]

[...] nor Ubuntu, [..] have boot media with 6.13 yet.

Yes, the newest Ubuntu 24.10 has 6.11 ... but ... Ubuntu do a lot of patches for its kernel ... so there is a chance (but I dont know) that the needed ethernet module is in there ... just try it

[xevra]

[...] nor Ubuntu, [..] have boot media with 6.13 yet.

Yes, the newest Ubuntu 24.10 has 6.11 ... but ... Ubuntu do a lot of patches for its kernel ... so there is a chance (but I dont know) that the needed ethernet module is in there ... just try it

My friends, we have internet.

We still don't have ethernet, but we have wifi, which feels backwards to me, but I'll take it!

[xevra]

I have managed to encounter the same kernel panic while trying to web install.

I have about a dozen other things I want to try now, but that will be tomorrow.

We have wifi and sshd. No more sneakernet. That is progress.

[pietinger]

[...] We still don't have ethernet, but we have wifi, which feels backwards to me, but I'll take it!

@Neddy: @xevra is using our Gentoo-distribution-kernel ... which has everything which is necessary for an initramfs ... most important is in grub.cfg the line:

Code: Select all

initrd   /boot/amd-uc.img /boot/initramfs-6.13.4-gentoo-dist.img

below of

Code: Select all

linux   /boot/vmlinuz-6.13.4-gentoo-dist root=UUID=....

So, for me it looks correct. ( /boot/amd-uc.img is the CPU microcode for your AMD; u means µ)

[NeddySeagoon]

pietinger,

Yes, it all looks correct but it doesn't work, so its not. :)

xevra,

You may have two different places on your system called boot. boot is a partition /boot (leading /) is a directory on the root filesystem.
For updating the kernel, the boot partition is normally mounted at /boot, so they look to be the same.
When you forget to mount the boot partition on the /boot mount point, everything goes into the /boot directory on the root partition.
That's not an error ... yet.

When grub is doing its thing, the kernel filesystem tree does not exist. It has to make its own arrangements to read the files it needs.
It cannot call the kernel as the kernel is not yen loaded. Grub is going to do that.

We can tell from

Code: Select all

	search --no-floppy --fs-uuid --set=root 4643c03f-299a-4e28-b975-5c95b56244f0
	echo	'Loading Linux 6.13.4-gentoo-dist ...'
	linux	/boot/vmlinuz-6.13.4-gentoo-dist root=UUID=4643c03f-299a-4e28-b975-5c95b56244f0 ro net.ifnames=0 
	echo	'Loading initial ramdisk ...'
	initrd	/boot/amd-uc.img /boot/initramfs-6.13.4-gentoo-dist.img

That you do indeed have a separate boot partition. Its UUID is 4643c03f-299a-4e28-b975-5c95b56244f0 which is different to root=UUID=4643c03f-299a-4e28-b975-5c95b56244f0

After all that, please check the following ...
With the boot partition mounted at /boot

Code: Select all

ls -l /boot 

should show lots of files.
With the boot partition not mounted,

Code: Select all

ls -l /boot

should show nothing. If there are files there, it shows that you forgot to mount the boot partition at some time. Grub will not find them.

[pietinger]

[...] That you do indeed have a separate boot partition. [...]

I am sorry, I have not seen, I have only looked at his installation routine description:

Code: Select all

# Preparing the disks
mkfs.vfat -F 32 /dev/sda1
mkswap /dev/sda2
mkfs.xfs /dev/sda3
swapon /dev/sda2
mount /dev/sda3 /mnt/gentoo/
mkdir /mnt/gentoo/efi
mount /dev/sda1 /mnt/gentoo/efi

[xevra]

Okay, I have more things to try, but I was able to boot the stable kernel from a web install using dracut.

Still no ethernet, and I haven't actually configured wpa-supplicant yet, and my whole screen is yellow, but dhcpcd does find my wifi card.

Have some errands to run, but we have achieved first boot.

[NeddySeagoon]

pietinger,

boot and the ESP may be same partition. I can't tell.
grub.cfg tells that gpubs root is not the same partition as the kernels root.

All the boot files in the ESP works, as long as no symlinks are required.

[xevra]

The yellow screen was fixed by installing x11-drivers/nvidia-drivers

[xevra]

wpa-supplicant is now configured (although the ip keeps jumping around because I suspect NetworkManager and wpa_supplicant are fighting for control of the device, but I will remedy this is in the next installation by just installing wpa_supplicant on its own).

I have successfully installed neofetch over the wifi:

Code: Select all

         -/oyddmdhs+:.                xevra@Pharaon 
     -odNMMMMMMMMNNmhy+-`             ------------- 
   -yNMMMMMMMMMMMNNNmmdhy+-           OS: Gentoo Linux x86_64 
 `omMMMMMMMMMMMMNmdmmmmddhhy/`        Host: X870 AORUS ELITE WIFI7 ICE -CF-WCP 
 omMMMMMMMMMMMNhhyyyohmdddhhhdo`      Kernel: 6.12.16-gentoo-dist 
.ydMMMMMMMMMMdhs++so/smdddhhhhdm+`    Uptime: 10 mins 
 oyhdmNMMMMMMMNdyooydmddddhhhhyhNd.   Packages: 442 (emerge) 
  :oyhhdNNMMMMMMMNNNmmdddhhhhhyymMh   Shell: bash 5.2.37 
    .:+sydNMMMMMNNNmmmdddhhhhhhmMmy   Terminal: /dev/pts/0 
       /mMMMMMMNNNmmmdddhhhhhmMNhs:   CPU: AMD Ryzen 9 9900X (24) @ 5.658GHz 
    `oNMMMMMMMNNNmmmddddhhdmMNhs+`    GPU: NVIDIA TITAN V 
  `sNMMMMMMMMNNNmmmdddddmNMmhs/.      GPU: AMD ATI Radeon Graphics 
 /NMMMMMMMMNNNNmmmdddmNMNdso:`        Memory: 606MiB / 61871MiB 
+MMMMMMMNNNNNmmmmdmNMNdso/-
yMMNNNNNNNmmmmmNNMmhs+/-`                                     
/hMMNNNNNNNNMNdhs++/-`                                        
`/ohdmmddhys+++/:.`
  `-//////:--.

Thank you all for helping me get this far. At this point I have a functional computer, and the wifi card was one of the key selling points of this motherboard, so I could make do.

However, the scientist in me wants to continue debugging the ethernet drivers. If you would like to keep investigating alongside me, I would appreciate your help.

I think the issues I was having before were caused by one of three things:
- -march=native with hardware that is too new
- ugrd
- 6.13 itself.
All three of these things worked together on my other computer, but this combination didn't work on my new computer.
I'm going to try turning them back on one at a time.

Neddy,

Even on the successfully booted install, there is no "/usr/src/linux/Documentation/admin-guide" folder. I suspect that it is removed from the version of the linux kernel repository which is shipped as part of the distribution kernel.

That you do indeed have a separate boot partition. Its UUID is 4643c03f-299a-4e28-b975-5c95b56244f0 which is different to root=UUID=4643c03f-299a-4e28-b975-5c95b56244f0

Even a CRTL-F reveals that these are the same devices. I suspect that this is something funny about the distribution kernel.

I can confirm that the SSD only has the three partitions.(/efi, swap, and /)

Thanks

[NeddySeagoon]

xevra,

You can have wpa-supplicant and NetworkManager installed but you must only start NetworkManager. It will use wpa-supplicant.

Your wired network needs the 6.13 kernel but you have 6.12.
There is no need for a reinstall.

You can add the 6.13 kernel to your install and choose which one to boot from the grub menu.
If 6.13 is still broken, it a menu choice to go back to 6.12

[xevra]

xevra,

You can have wpa-supplicant and NetworkManager installed but you must only start NetworkManager. It will use wpa-supplicant.

Your wired network needs the 6.13 kernel but you have 6.12.
There is no need for a reinstall.

You can add the 6.13 kernel to your install and choose which one to boot from the grub menu.
If 6.13 is still broken, it a menu choice to go back to 6.12

NeddySeagoon,

It was just as you said.

I am now running 6.13 and my ethernet device is working.

Thank you!